New robust point registration algorithm by maximization of entropy and mutual information

Point registration is an important problem in many fields, such as machine vision, medical image registration. A new robust point registration algorithm is proposed by maximization of entropy and mutual information which is very accurate and rapid. A novel aspect of the algorithm is the introduction of mutual information into point registration algorithm, and another aspect is the joint probability matrix, which can represent the correspondences of two point-sets. Then a new energy function, including the joint probability matrix and spatial mapping matrix, has been derived. By minimizing the above energy through a deterministic annealing algorithm, the estimation of spatial mapping parameters and joint probability matrix can be obtained. The experimental results show the validation and robustness of the algorithm.